From 6391d404a69385cb67b3725c3695d5f4e177e4a4 Mon Sep 17 00:00:00 2001 From: Michael Witrant Date: Sat, 24 Sep 2011 12:41:52 +0200 Subject: [PATCH] fix language --- consultation_ipred/organisations.yaml | 18 ++++++------ consultation_ipred/parser.rb | 18 ++++++++++-- consultation_ipred/public_authorities.yaml | 32 +++++++++++----------- 3 files changed, 40 insertions(+), 28 deletions(-) diff --git a/consultation_ipred/organisations.yaml b/consultation_ipred/organisations.yaml index eac8837..bf4a269 100644 --- a/consultation_ipred/organisations.yaml +++ b/consultation_ipred/organisations.yaml @@ -48,14 +48,6 @@ European Fashion Council: !ruby/struct:Answer languages: - en -"": !ruby/struct:Answer - name: "" - files: - - https://circabc.europa.eu/d/d/workspace/SpacesStore/318e33ff-694a-4485-9755-dde97697d148/fast.pdf - annexes: [] - - languages: - - fast Voez: !ruby/struct:Answer name: Voez files: @@ -338,7 +330,7 @@ Aepoc: !ruby/struct:Answer annexes: [] languages: - - " en" + - en Rim: !ruby/struct:Answer name: Rim files: @@ -676,6 +668,14 @@ Faep: !ruby/struct:Answer languages: - en +Fast: !ruby/struct:Answer + name: Fast + files: + - https://circabc.europa.eu/d/d/workspace/SpacesStore/318e33ff-694a-4485-9755-dde97697d148/fast.pdf + annexes: [] + + languages: [] + Danish Patent Agents Association: !ruby/struct:Answer name: Danish Patent Agents Association files: diff --git a/consultation_ipred/parser.rb b/consultation_ipred/parser.rb index a41ffb3..e0e2436 100644 --- a/consultation_ipred/parser.rb +++ b/consultation_ipred/parser.rb @@ -20,13 +20,23 @@ files.each do |type| name = CGI::unescape(name) #name = Iconv.iconv("utf-8", "iso8859-15", name).first names = name.split("_") - language = names.pop + + language = names.last.dup + + language.gsub!(/\s/, "") + if language.size == 2 + names.pop + else + language = nil + end annex = names.index { |part| part =~ /^annex/ } if annex names[annex..-1] = [] end + raise "no name on #{url}" if names.empty? + name = names.map(&:capitalize).join(" ") #name = "#{name} (#{language})" @@ -36,8 +46,10 @@ files.each do |type| else answer.files << url end - answer.languages << language - answer.languages.uniq! + if language + answer.languages << language + answer.languages.uniq! + end end answers.sort.each do |name, answer| diff --git a/consultation_ipred/public_authorities.yaml b/consultation_ipred/public_authorities.yaml index 9a667e9..f881a13 100644 --- a/consultation_ipred/public_authorities.yaml +++ b/consultation_ipred/public_authorities.yaml @@ -1,20 +1,4 @@ --- -Uk: !ruby/struct:Answer - name: Uk - files: - - https://circabc.europa.eu/d/d/workspace/SpacesStore/6436e9a1-1ba6-414e-af8c-d50fa3ac2c52/uk_governement.pdf - annexes: [] - - languages: - - governement -"": !ruby/struct:Answer - name: "" - files: - - https://circabc.europa.eu/d/d/workspace/SpacesStore/4edaf82a-d143-4a07-9880-e59529afc595/france.pdf - annexes: [] - - languages: - - france Portugal: !ruby/struct:Answer name: Portugal files: @@ -105,6 +89,14 @@ Italy Office Of The Prime Minister: !ruby/struct:Answer languages: - it +France: !ruby/struct:Answer + name: France + files: + - https://circabc.europa.eu/d/d/workspace/SpacesStore/4edaf82a-d143-4a07-9880-e59529afc595/france.pdf + annexes: [] + + languages: [] + Finnish Commerce Federation: !ruby/struct:Answer name: Finnish Commerce Federation files: @@ -212,6 +204,14 @@ Czech Republic: !ruby/struct:Answer languages: - en +Uk Governement: !ruby/struct:Answer + name: Uk Governement + files: + - https://circabc.europa.eu/d/d/workspace/SpacesStore/6436e9a1-1ba6-414e-af8c-d50fa3ac2c52/uk_governement.pdf + annexes: [] + + languages: [] + Germany: !ruby/struct:Answer name: Germany files: -- GitLab